Google AMP: Cố gắng sửa chữa web bằng cách kiểm soát nó
Vào năm 2015, Google đã có kế hoạch để cứu vãn web di động bằng cách kiểm soát nó. Và trong một thời gian ngắn, ngành truyền thông thực sự không có sự lựa chọn nào khác ngoài việc phải tham gia.
Bắt đầu vào một buổi sáng đầy vui vẻ vào tháng Mười ở New York, công ty đã tập hợp báo chí tại một quán ăn sáng tên là Sadelle’s ở SoHo. Khi các phóng viên được tụ hợp tại đây để ăn bánh mì và cá hồi, Phó chủ tịch phụ trách tin tức của Google, Richard Gingras, đã giải thích rằng web mở đang trong cuộc khủng hoảng. Trang web quá chậm, quá khó sử dụng và quá nhiều quảng cáo. Kết quả là, ông cảnh báo rằng mọi người đang đổ về những trải nghiệm tốt hơn được cung cấp bởi các nền tảng xã hội và các cửa hàng ứng dụng. Nếu xu hướng này tiếp tục, đó sẽ là sự kết thúc của web như chúng ta biết.
Nhưng Google đã có một kế hoạch để chiến đấu trở lại: Trang web di động được tăng tốc, hoặc AMP, một định dạng mới để thiết kế trang web dành cho thiết bị di động. AMP sẽ đảm bảo rằng web di động có thể nhanh hơn, dễ sử dụng hơn, tải lên ngay lập tức và không kém phần được ưa chuộng như ứng dụng di động. “Chúng tôi có tại đây để đảm bảo rằng web tiếp tục phát triển và toàn bộ trọng tâm của chúng tôi là vào nỗ lực đó,” Gingras nói. “Chúng tôi có tại đây để làm cho web trở nên tuyệt vời hơn.”
“Làm cho web trở nên tuyệt vời hơn” là một thuật ngữ phổ biến trong Google vào thời điểm đó, phản ánh chiến dịch tổng thống mới nổi của một ứng cử viên đảng Cộng hòa là Donald Trump. Có rất nhiều công việc kỹ thuật đằng sau khẩu hiệu này: Google đang xây dựng trình duyệt Chrome của riêng mình thành một hệ thống hoạt động đầu tiên dành cho máy tính xách tay; cố gắng thay thế các ứng dụng bản địa bằng Progressive Web Apps; thúc đẩy việc áp dụng chuẩn HTTPS an toàn hơn trên web; và quảng bá các tên miền cấp cao mới sẽ nhằm làm cho .blog và .pizza quan trọng như .com. Rất nhiều trong đó là nhàm chán hoặc khó hiểu đối với các nhà quản lý truyền thông. Điểm chính là Google cam kết giành lại quyền phân phối từ Apple và Facebook và trao nó cho các nhà xuất bản.
“Sau mười năm, khi các tờ báo biến mất, lưu lượng đọc tạp chí thu nhỏ và doanh nghiệp trang web suy giảm, ngành truyền thông đã trở nên bất lực. Ngay cả những nhà xuất bản hay nghi ngờ nhất cũng đã quen với việc chơi bất kỳ trò chơi nào mà các nền tảng như Google và Facebook đòi hỏi để tìm kiếm lưu lượng truy cập. Và khi Facebook chuyển hướng một cách hỗn loạn sang video, đó để lại Google là tài xế lớn nhất của lưu lượng truy cập đến các trang web trên toàn mạng. Ai có lựa chọn nào khác?” ông Terence Eden, một chuyên gia chuẩn web và một thành viên trước của Hội đồng tư vấn Google AMP, cho biết. Một nhà điều hành truyền thông đã làm việc trên các dự án AMP nhưng yêu cầu giấu tên giống như các nguồn khác trong câu chuyện này, khung bối trao đổi thậm chí còn đơn giản hơn: “Nếu bạn muốn tiếp cận với đối tượng người xem này, bạn cần phải tuân thủ các quy tắc này.”
Việc áp dụng phiên bản mới lạ của Google về web đã dẫn đến một dòng lưu lượng truy cập không thể cưỡng lại cho các nhà xuất bản lúc đầu: sử dụng AMP đã tăng lưu lượng tìm kiếm đến một trang web của một tạp chí quốc gia lớn lên đến 20%, theo nhà điều hành thực hiện việc triển khai.
Tuy nhiên, AMP đi kèm với những hậu quả lớn, đặc biệt là trong việc xác định cách các trang web được tiếp thị. AMP làm cho việc sử dụng công nghệ quảng cáo không phải từ Google trở nên khó khăn hơn, gãy đứt mối quan hệ giữa Google và truyền thông đến mức AMP trở thành một thành phần chính trong một vụ kiện chống độc quyền được khởi kiện chỉ 5 năm sau khi ra mắt vào năm 2020 bởi 17 luật sư tổng giám đốc bang, cáo buộc Google đang duy trì một độc quyền bất hợp pháp trong ngành quảng cáo. Các bang này cho rằng Google thiết kế AMP, một phần là để ngăn chặn các nhà xuất bản sử dụng các công cụ quảng cáo thay thế – công cụ sẽ tạo ra nhiều tiền hơn cho các nhà xuất bản và ít hơn cho Google. Một vụ kiện khác được khởi kiện vào tháng 1 năm 2023 bởi Bộ Tư pháp Hoa Kỳ đã đi xa hơn nữa, cáo buộc Google đã lấy AMP làm “nỗ lực để đẩy một phần của web mở vào một vườn tường kiểm soát bởi Google, nơi Google có thể ra lệnh trực tiếp hơn về cách khối quảng cáo số có thể được bán.”
Tại đây vào năm 2023, AMP dường như đã phai nhạt. Hầu hết các nhà xuất bản đã bắt đầu từ bỏ hỗ trợ và thậm chí Google cũng không quan tâm nhiều hơn nữa. Sự trỗi dậy của ChatGPT và các dịch vụ trí tuệ nhân tạo khác đặt ra mối đe dọa trực tiếp hơn cho hoạt động tìm kiếm của Google hơn bao giờ hết. Tuy nhiên, ngành truyền thông vẫn phụ thuộc vào dòng lưu lượng truy cập từ Google và khi công ty tìm kiếm động thái tiếp theo, câu chuyện về cách nó dùng AMP một cách tàn nhẫn để kiểm soát cấu trúc và hoạt động của web chỉ rõ cách mà Google sẽ đi để bảo vệ hoạt động của mình – và web có thể bất lực để ngăn chặn nó. AMP đã thành công ngoạn mục. Sau đó thất bại. Và với bất kỳ ai đang tìm kiếm lý do để không tin tưởng vào công ty lớn nhất trên Internet này, câu chuyện về AMP chứa đựng tất cả những bằng chứng mà bạn cần.
#Google #AMP #Web #Công nghệ #Truyền thông #Lưu lượng truy cập #Trang web #Xây dựng #Quảng cáo #Kiểm
In 2015, Google hatched a plan to save the mobile web by effectively taking it over. And for a while, the media industry had practically no choice but to play along.
It began on a cheery October morning in New York City; the company had gathered the press together at a buzzy breakfast spot named Sadelle’s in SoHo. As the assembled reporters ate their bagels and lox, Google’s vice president of news, Richard Gingras, explained that the open web was in crisis. Sites were too slow, too hard to use, too filled with ads. As a result, he warned, people were flocking to the better experiences offered by social platforms and app stores. If this trend continued, it would be the end of the web as we know it.
But Google had a plan to fight back: Accelerated Mobile Pages, or AMP, a new format for designing mobile-first webpages. AMP would ensure that the mobile web could be as fast, as usable, instantly loading, and every bit as popular as mobile apps. “We are here to make sure that the web evolves, and our entire focus is on that effort,” Gingras said. “We are here to make the web great again.”
“Make the web great again” was a popular phrase across Google at the time, echoing the burgeoning presidential campaign of an upstart Republican named Donald Trump. There was a lot of technical work behind the slogan: Google was building its own Chrome browser into a viable web-first operating system for laptops; trying to replace native apps with Progressive Web Apps; pushing to make the more secure HTTPS standard across the web; and promoting new top-level domains that would aim to make .blog and .pizza as important as .com. Much of this was boring or went over the heads of media execs. The point was that Google was promising to wrest distribution power away from Apple and Facebook and back into the hands of publishers.
“You want access to this audience, you need to play by these rules.”
After a decade of newspapers disappearing, magazine circulations shrinking, and websites’ business dwindling, the media industry had become resigned to its own powerlessness. Even the most cynical publishers had grown used to playing whatever games platforms like Google and Facebook demanded in a quest for traffic. And as Facebook chaotically pivoted to video, that left Google as the overwhelming driver of traffic to websites all over the web. What choice did anyone have?
“If Google said, ‘you must have your homepage colored bright pink on Tuesdays to be the result in Google,’ everybody would do it, because that’s what they need to do to survive,” says Terence Eden, a web standards expert and a former member of the Google AMP Advisory Committee. One media executive who worked on AMP projects but who, like other sources in this story, requested anonymity to speak about Google, framed the tradeoff even more simply: “you want access to this audience, you need to play by these rules.”
Adopting Google’s strange new version of the web resulted in an irresistible flood of traffic for publishers at first: using AMP increased search traffic to one major national magazine’s site by 20 percent, according to the executive who oversaw the implementation.
But AMP came with huge tradeoffs, most notably around how all those webpages were monetized. AMP made it harder to use ad tech that didn’t come from Google, fraying the relationship between Google and the media so badly that AMP became a key component in an antitrust lawsuit filed just five years after its launch in 2020 by 17 state attorneys general, accusing Google of maintaining an illegal monopoly on the advertising industry. The states argue that Google designed AMP in part to thwart publishers from using alternative ad tools — tools that would have generated more money for publishers and less for Google. Another lawsuit, filed in January 2023 by the US Justice Department, went even further, alleging that Google envisioned AMP as “an effort to push parts of the open web into a Google-controlled walled garden, one where Google could dictate more directly how digital advertising space could be sold.”
Here in 2023, AMP seems to have faded away. Most publishers have started dropping support, and even Google doesn’t seem to care much anymore. The rise of ChatGPT and other AI services pose a much more direct threat to its search business than Facebook Instant Articles and Apple News ever did. But the media industry is still dependent on Google’s fire hose of traffic, and as the company searches for its next move, the story of how it ruthlessly used AMP in an attempt to control the very structure and business of the web makes clear exactly how far it will go to preserve its business — and how powerless the web may be to stop it.
AMP succeeded spectacularly. Then it failed. And to anyone looking for a reason not to trust the biggest company on the internet, AMP’s story contains all the evidence you’ll ever need.
The small-screen shake-up
Earlier in 2015, months before AMP launched, one of Google’s key metrics was on the verge of a dramatic flip: the volume of searches coming from mobile phones was just about to outnumber the ones coming from desktop and laptop computers. This shift had been a long time coming, and Google saw it as an existential threat. The company had become a nearly $75 billion annual business almost entirely on ads — which made up about 90 percent of its revenue — and the most important ones by far were the ones atop search results in desktop browsers. By some internal measures, a typical mobile search at the time brought in about one-sixth as much ad revenue as on desktop. The increasingly mobile-focused future could mean a disastrous revenue drop for Google.
In public, Google framed AMP as something like a civic mission, an attempt to keep the web open and accessible to everyone instead of moving to closed gardens like Facebook Instant Articles or Apple News, which offered superior mobile reading experiences. “To some degree, on mobile, (the web) has not fully satisfied users’ expectations,” Gingras said at the launch event. “We are hoping to change that.”
But the fight to fix the mobile web wasn’t just an altruistic move in the name of teamwork and openness and kumbaya. Internally, some viewed it as a battle for Google’s own survival. As smartphones became the default browsing experience for billions of users around the world, the mobile web was becoming the only web that really mattered. Google’s competitors were exerting far more control over how users lived their lives on their phones: readers were getting their news from native apps and from proprietary formats created by Facebook and Apple. Google worried that if enough users switched to these faster, simpler, more controlled experiences, it risked being left out altogether.
As Big Tech companies took over the ad industry, it did so largely at the expense of publishers. Newspapers used to be the way to advertise your new hair salon, or you might buy local TV ads to hawk the latest appliances for sale in your store. By 2015, most advertisers just went through Facebook and Google, which offered a more targeted and more efficient way to reach buyers.
“You meet with a Facebook person and you see in their eyes they’re psychotic.”
Google, obviously aware that it was taking revenue from publishers, occasionally tried to make nice. Sometimes that meant creating new products, like the awkwardly named Google Play Newsstand, to give media companies another place to distribute and sell content. Sometimes — often, actually — it meant just giving publishers a bunch of money whenever a government would get mad, like the €60 million “Digital Publishing Innovation Fund” Google set up in France after a group of European publishers sued and settled with the search giant.
This “we care about publishers!” dance is a staple of Silicon Valley. Apple briefly promised to save the news business with the iPad, convincing publishers around the world to build bespoke tablet magazines before mostly abandoning that project. Facebook remains in a perpetually whipsawing relationship with the media, too: it will promote stories in the News Feed only to later demote them in favor of “Meaningful Social Interactions,” then promise publishers endless video eyeballs before mostly giving up on Facebook Watch.
The platforms need content to keep users entertained and engaged; publishers need distribution for their content to be seen. At best, it’s a perfectly symbiotic relationship. At worst, and all too often, the platforms simply cajole publishers into doing whatever the platforms need to increase engagement that quarter.
For publishers over the last decade, chasing platform policies and supporting new products has become the only means of survival. “That’s the sort of tradeoff publishers are used to,” says one media executive who was involved with AMP in its early days. “Do it this way and you’ll get an audience.” But while publishers had long been wary of the tendency of Big Tech companies to suck up ad dollars and user data, they had seen Google as something closer to a partner. “You meet with a Facebook person and you see in their eyes they’re psychotic,” says one media executive who’s dealt with all the major platforms. “The Apple person kind of listens but then does what it wants to do. The Google person honestly thinks what they’re doing is the best thing.”
As Facebook grew, and social media in general began to replace blogs and forums, it felt like Google’s view of the internet was shrinking
Phones potentially made all of this harder. For Google, search was harder to monetize on smaller screens with correspondingly fewer ad slots, and it was also, in some ways, an inferior product. That was largely for reasons out of Google’s control: many of the mobile websites Google sent users to were slow, covered in autoplaying video and unclosable ads, and generally considered a worse experience than the apps that publishers and media organizations had been focused on for the last several years. Google executives talked often internally about being ashamed of sending people to some websites.
But the big reason for consternation within Google was a company just a few miles down the road. If mobile was going to win, then so was Facebook. This was pre-metaverse Facebook, of course, when the company was a booming social networking giant, a thriving ad business, and a mobile success story: Facebook reported in April 2015 that it had 1.25 billion mobile active users on its products every month and that nearly three-quarters of its advertising revenue was coming from mobile.
Facebook was, to most users, a mobile app, not a website. Google can’t crawl a mobile app. And it got worse: most content on Facebook was shared among friends and followers and, as such, was completely opaque to Google, even on the web. For most of its existence, Google could take for granted that the vast majority of the internet’s content would be open and searchable. As Facebook grew, and social media in general began to replace blogs and forums, it felt like Google’s view of the internet was shrinking.
Meanwhile, Mark Zuckerberg made no secret of Facebook’s ambitions to take on Google, to take on everybody, really: the CEO’s aim was to turn Facebook into a platform the size of the internet. But he wanted to win at search, too, first by better indexing Facebook content and then by ultimately doing the same to the web. “There’s a lot of public content that’s out there that any web search engine can go index and provide,” he told investors in the spring of 2015.
“They said they can’t fix the ads. It’s too hard.”
The simplest thing to do would be to beat Facebook at its own game. But Google had already tried that — a few times. Seeing the rise of social networking, and the threat that friend-sourced content posed to Google’s search-based business model, the company poured resources into the Google Plus social network. But it never caught on and, by 2015, was effectively on its last legs. There was simply no way to out-Facebook Facebook.
Around the same time, Facebook also launched Instant Articles, a Facebook-specific tool that turned web articles into native posts on the platform. The pitch for Instant Articles was simple: they would speed up the News Feed, making it quicker to read stories so users didn’t have to suffer through the mobile web’s interminable load times and hideous pages. Instant Articles made some publishers nervous since it effectively loaded their content directly onto Facebook’s platform and gave the company complete control over their audiences. Some opted out entirely. But many others saw too big a potential audience to ignore and developed tools to syndicate their stories as Instant Articles.
A few months later, Apple launched Apple News, its own proprietary article format and app for displaying publisher content. At its own developer conference that spring, Apple’s then-VP of product management, Susan Prescott, made a case that sounded eerily like Facebook’s. “The articles can come from anywhere,” she said, “but the best ones are built in our new Apple News format.” Software chief Craig Federighi followed up with a backhanded swipe at Google News and Facebook. “Unlike just about any other news aggregation service we’re aware of on the planet, News is designed from the ground up with your privacy in mind.”
The media industry, collectively, bought the hype around what came to be known as “distributed publishing.” “Is the media becoming a wire service?” asked Ezra Klein at Vox in a piece that kicked off a million AMP and Instant Articles projects. “My guess is that within three years, it will be normal for news organizations of even modest scale to be publishing to some combination of their own websites, a separate mobile app, Facebook Instant Articles, Apple News, Snapchat, RSS, Facebook Video, Twitter Video, YouTube, Flipboard, and at least one or two major players yet to be named,” he wrote. “The biggest publishers will be publishing to all of these simultaneously.”
To some at Google, all of this looked a lot like a few proprietary platforms conspiring to kill the open web. Which might kill Google. Search — and its behemoth ad business — only worked if the web was full of open, indexable pages that its search crawlers could see and direct users to. Instant Articles and Apple News also gave those platforms control over the advertising on their pages, which threatened AdWords, another of Google’s largest revenue streams.
Phone browsers were bad; the webpages were even worse
Over the course of 2015, as Google debated internally how best to respond, the company also hosted a clubby “unconference” called Newsgeist. Google held these periodically in partnership with the Knight Foundation as a way to work with and hear from the news industry. Jeff Jarvis, a CUNY professor and media critic, had been agitating at Newsgeist events for years for Google to build what he called “the embeddable newspaper,” a way for news articles to be displayed around the internet in much the same way a YouTube video can be embedded practically anywhere. Gingras also liked the idea; he was a big believer in what he called “portable content.”
In May 2015, at the first Newsgeist Europe in Helsinki, Finland, Instant Articles was a topic of much conversation. Jarvis, in particular, saw Instant Articles as a useful technical prototype with all the wrong attributes: it was closed off, only worked on one platform, and accrued no value back to publishers. Jarvis spent time at the conference arguing for someone — presumably Google — to build a better alternative.
Ultimately, what the company built was AMP. Done right, it could bring the same speed, simplicity, and design to the entire internet — without closing it off. To lead the effort, Google designated two people who had come from Google Plus: David Besbris, who had led the company’s wayward social networking effort, and Malte Ubl, who helped to build the social network’s technical infrastructure.
At least, that’s how Google described it publicly. According to interviews with former employees, publishing executives, and experts associated with the early days of AMP, while it was waxing poetic about the value and future of the open web, Google was privately urging publishers into handing over near-total control of how their articles worked and looked and monetized. And it was wielding the web’s most powerful real estate — the top of search results — to get its way.
“(Google) came to us and said, the internet is broken, ads aren’t loading, blah blah, blah. We want to provide a better user experience to users by coming up with this clean standard,” says one magazine product executive. “My reaction was that the main problem is ads, so why don’t you fix the ads? They said they can’t fix the ads. It’s too hard.”
Faster, faster, faster
Before it was called AMP, Google’s nascent web standard was known as PCU — Portable Content Unit. The team of Googlers building the new format had only one goal, or at least only one that mattered: make webpages faster. There were lots of other goals, like giving publishers monetization and branding options, but all of that was secondary to load times. If the page appeared instantly after a user tapped the link in search results, AMP would feel as instant and native as an app. Nothing else mattered as much as speed.
Google had tried in the past to incentivize publishers to make their own webpages faster. Load times had long been a factor in how the search engine ranked sites on desktop, for instance, and load times were presented front and center in Google Analytics. Google even built a tool called “Instant Pages” that tried to guess which sites users would click on and pre-render those pages so they’d appear more quickly.
And yet, the mobile web still, in a word, sucked. “Publishers, frankly, then — and to a great degree still now — considered mobile web traffic to be essentially junk traffic,” says Aron Pilhofer, a longtime media executive and now a journalism professor at Temple University. Many mobile websites were completely separate entities from their desktop pages, prefaced with “mobile.” or “m.” in their URLs. Publishers compensated for small screens with more ads per page, and the whole industry was in the midst of an unfortunate obsession with autoplaying video. Phone browsers were bad; the webpages were even worse.
Google didn’t have great tools for understanding mobile pages at the time, so it couldn’t easily issue the same “we just like fast pages” edict. It could take the effort to develop those metrics and then urge publishers to update their sites to meet Google’s bar for speed, but there simply wasn’t time. Internally, Google felt it needed a solution immediately. Competition was here. AMP was a blunt object, but it was designed to get results quickly. AMP’s purpose, Google’s Gingras said at the 2015 launch event, “is about making sure the World Wide Web is not the World Wide Wait.”
AMP was, in many ways, a step backward for the web. Nieman Lab’s Joshua Benton noted at the time that Google’s sample AMP-powered webpages “look a lot like the web of, say, 2002, shrunk down to a phone screen.”
But it was fast. And to Google, that was all that mattered.
The growth hack to end all growth hacks
For AMP to work, Google knew it needed to get broad adoption. But simply asking publishers to support a new standard wouldn’t be easy. Publishers were already neglecting their mobile websites, which was the whole problem, and they weren’t likely to sign up to work on them just for Google’s benefit.
The team tried a few things to get more AMP content, like auto-converting stories from the Google Play Newsstand and elsewhere. WordPress began working on a plug-in that made creating AMP pages as easy as checking a box every time you published a post. One way some people in and outside of Google thought of AMP was similar to RSS — another syndication format, another box to click next to the one that tweets the story and posts the top image on Instagram. But Google worried that this approach would give all AMP pages a same-y, boring look and reader experience. What Google really needed was for publishers to not just support AMP but also embrace it.
The team quickly landed on a much more powerful growth hack: Google’s search results. It would be easy for Google to factor AMP into the way it ranks search results, to effectively tell publishers that AMP-powered pages would be higher on the list, and anything else would be pushed down the page. (It had previously done something similar with HTTPS, another push toward a new web standard.) Publishers, most of them existentially reliant on the fire hose of Google traffic, would have no choice but to give in and use AMP.
Such an aggressive move would be a bad look for Google, though, not to mention a potentially anti-competitive one, especially given that the company has always maintained it cares about a webpage’s “relevance” above all else. But there was a middle ground, or maybe a loophole: a relatively new product in Google search known as the Top Stories carousel, which showed a handful of horizontally scrolling news stories at the top of some search results pages. They weren’t part of the search results, the “10 blue links” Google is known for and so scrutinized over. They were something separate, so the rules could be different.
“It felt faster because Google cheated.”
Google said from the beginning that AMP would not be a factor in regular “10 blue link” search results. (Several publishing executives say they’re still not sure if that was true: “when Google said AMP doesn’t matter, no one believes them,” one says. The company denies that it has ever been a factor in search result rankings.) But only AMP pages would be included in the carousel, with a lightning bolt in the corner to signify that tapping that card would offer the instant loading experience users were getting from Instant Articles and Apple News.
That carousel took up most of the precious space on a phone screen, which made Top Stories some of the most important real estate on the mobile web. And so, the growth hack worked. When AMP launched in early 2016, a who’s who of publishers had signed up to support the new format: The Guardian, The Washington Post, BuzzFeed, the BBC, The New York Times, and Vox Media, The Verge’s parent company, all quickly began developing for AMP. Others would join in the months that followed.
But many of those publishers weren’t necessarily signing up because they believed in AMP’s vision or loved the tech. Far from it. Google’s relentless focus on page speed, and on shipping as quickly as possible to thwart Facebook and Apple, meant the first versions of AMP couldn’t do very much. It didn’t support comments or paywalls, and the restrictions on JavaScript meant publishers couldn’t bring in third-party analytics or advertising. Interactive elements, even simple things like tables and charts, mostly didn’t work.
AMP, it turned out, wasn’t even that fast. Multiple publishers ran internal tests and found they were able to make pages that loaded more quickly than AMP pages, so long as they were able to rein in the ad load and extra trackers. It was much harder to build slow pages on AMP — in part because AMP couldn’t do very much — but there were lots of other ways to build good pages.
“We built a standard, it’s shit, it’s terrible, it’s not ready, it does only like a quarter of what you need it to do.”
And even if AMP pages did seem to load faster from search results, “it felt faster because Google cheated,” says Barry Adams, a longtime SEO consultant. When publishers built AMP-powered pages, they submitted them to Google’s AMP Validator, which made sure the page worked right — and cleared it for access to the carousel. As it was checking the code, Google would grab a copy of the entire page and store it on Google’s own servers. Then, when someone clicked on the article in search results, rather than loading the webpage itself, Google would load its stored version. Any page pre-rendered like that would load faster, AMP or otherwise.
The AMP cache made it harder for publishers to quickly update their content — and made it nearly impossible for them to understand how people were using their sites. On cached pages, even the URL began with “google.com,” rather than the publisher’s own domain. It was as if Google had subsumed the entire publishing industry inside its office park in Mountain View.
Google kept promising publishers that this restrictive, Google-controlled version of AMP was just version one, that there was much more to come. But the carousel, that all-important new space in search results, required AMP from the beginning. “The problem was that when Google launched it, they also said, ‘You have to use AMP. We built a standard, it’s shit, it’s terrible, it’s not ready, it does only like a quarter of what you need it to do, but we need you to use it anyway because otherwise we’re just not going to show your articles in mobile search results anymore,’” Adams says. “And that is what ruffled everybody’s feathers.”
“The audience people hated it because it was against audience strategy,” says one former media executive who worked with AMP. “The data people hated it because it was against advertising and privacy strategy. The engineers hated it because it’s a horrendous format to work with… The analysts hated it because we got really bad behavioral data out of it. Everyone’s like, ‘Okay, so there’s no upside to this — apart from the traffic.’”
“Google’s strategy is always to create prisoner’s dilemmas that it controls — to create a system such that if only one person defects, then they win.”
On top of that, the traffic was worth less because it had fewer and more limited ads. “Every publisher experienced this — the AMP audience is less valuable. It’s millions of pennies and not having any dollars,” one executive says. “An AMP article earned 60 percent of what a (standard) article earned… It’s low enough to be noticeable. You were just playing the game of ‘if I didn’t have all this traffic, would I make more money?’”
“Google did not have an answer for the revenue gap — there was a lot of hand-waving, a lot of saying they would work with us,” says another executive. “Google on AMP was like Google on every product — lots of fanfare in the beginning, lots of grand plans, and then none of those plans ever saw the light of day.”
But the pageviews, in many cases, were enough to outweigh the costs. It’s almost impossible to overstate how important Google traffic is to most publishers. The analytics company Chartbeat estimated this year that search accounts for 19.3 percent of total traffic to websites, a number that doesn’t even include products like Google News and the news feed in the Google app, both of which also account for a huge portion of many publishers’ traffic. Google, as a whole, can account for up to 40 percent of traffic for even the largest sites. Disappearing from Google is life-and-death stuff.
Bigger media companies, those that could employ product and engineering staff of their own, could sometimes hack around AMP’s limitations — or, at the very least, deal with them without affecting the rest of the company’s business. Some big publishers came to see AMP as nothing more than some additional work required for a distributor. But even many smaller publishers, without the staff to manage the technical shortcomings or the resources to maintain yet another version of their website, still felt they had no choice but to support AMP.
As long as anyone played the game, everybody had to. “Google’s strategy is always to create prisoner’s dilemmas that it controls — to create a system such that if only one person defects, then they win,” a former media executive says. As long as anyone was willing to use AMP and get into that carousel, everyone else had to do the same or risk being left out.
Many within Google continued to see AMP as a net good, a way to make the web better and to keep it from collapsing into a few walled gardens. But to most publishers, AMP was, at best, just another app to send stuff to. “We didn’t see it as any different from building on Android or building on iOS,” one former media executive says. “It was this way to deliver the best mobile experience.” Supporting AMP was like supporting Apple News, Facebook Instant Articles, or even maintaining RSS feeds. It was just more work for more platforms.
That’s why the Top Stories carousel felt like a shakedown to so many publishers. Google claimed it was merely an incentive to do the obviously right thing and a nice boost in the user experience. But publishers sensed an unspoken message: comply with this new format or risk your precious search traffic. And your entire business.
Good governance
Despite all the issues with AMP’s tech and misgivings about Google’s intentions, the new format was a success from the very beginning. By December 2016, less than a year after its official launch, an Adobe study found that AMP pages already accounted for 7 percent of mobile traffic to “top publishers” in the US and grew 405 percent in just the final eight months of the year. Microsoft was planning to use AMP in the Bing app for iOS and Android. Twitter was looking into using it as well.
From the beginning, Google had proclaimed loudly that AMP was not a Google product. It was to be an open-source platform, all its source code available on GitHub for anyone to fork and edit and use to their own ends. AMP’s success was the web’s success, not Google’s.
In reality, Google exerted near-total authority over AMP. According to the 2020 antitrust lawsuit against Google, the company adopted a “Benevolent Dictator For Life” policy, and even when it transferred the AMP project to the OpenJS Foundation in 2019, it remained very much in charge. “When it suited them, it was open-source,” says Jeremy Keith, a web developer and a former member of AMP’s advisory council. “But whenever there were any questions about direction and control… it was Google’s.”
Several sources told me stories of heated arguments about the future of the web that ended in Google employees awkwardly reading lawyer-approved statements about things being open and opt in — and Google then getting its way. After a debate about the cache, and the data it gave Google, “they started bringing a whole bunch of people no one had ever heard of to committee meetings to say how wonderful the cache was,” one media exec remembers. And whenever there was debate about new features or the roadmap, Google always won.
“I have no doubt that the long-term play was to say, ‘We’re Google. This is a new language for the web. If you don’t like it, you’re not on the front page of Google anymore.’”
Over time, AMP began to support more ad networks — or, rather, more ad networks began to do the work required to support AMP’s locked-down structure. But many still felt the best experience was reserved for Google’s own ad tech. That fact has become the most contentious part of AMP’s history — and the reason it wound up in multiple antitrust lawsuits against Google. The suits allege, among other things, that Google used AMP as a way to curtail a practice called “header bidding,” which allows publishers to show their inventory to multiple ad exchanges at once in order to get the best price in real time. “Specifically,” the 2020 lawsuit says, “Google made AMP unable to execute JavaScript in the header, which frustrated publishers’ use of header bidding.” Google spokesperson Meghann Farnsworth said in a statement that “AG Paxton’s claims about AMP and header bidding are just false.” Most of the AMP-related provisions in that 2020 lawsuit were thrown out by a district court in 2022, which found that the case “does not plausible (sic) allege AMP to be an anticompetitive strategy.”
As AMP caught on, Google’s vision for the product became even more ambitious. The company started to suggest that, rather than maintain a website and a separate set of AMP pages, maybe some publishers should build their entire site within AMP. On launch day in October 2015, the AMP project website proudly proclaimed that it was “an architectural framework built for speed.” By the end of 2017, AMP was promising to enable “the creation of websites and ads that are consistently fast, beautiful and high-performing across devices and distribution platforms.” It was no longer just articles, and it was no longer just mobile. It was the whole web, rewritten Google’s way and forever compatible with its search engine.
“I 100 percent believe that Google would have loved to have said AMP is the future of HTML,” Eden says. “I have no doubt that the long-term play was to say, ‘We’re Google. This is a new language for the web. If you don’t like it, you’re not on the front page of Google anymore.’”
Ultimately, though, Google’s grandest ambitions didn’t come to pass. Neither did its smallest ambitions, really. As publishers continued to thrash against AMP’s constraints, and as overall scrutiny against Google ramped up, the company began to pull back.
The non-standard
In 2021, Google announced it would start featuring all pages in the Top Stories carousel, not just AMP-powered ones. Last May, Google let some local news providers for covid-related stories bypass this requirement. As soon as publishers didn’t have to use AMP anymore, they mostly stopped. The Washington Post abandoned it the same year, and a litany of others (including Vox Media) spent 2022 looking for ways off the platform. Even now, though, some of those publishers say they’re nervous about traffic disappearing. Google remains such a black box that it can be hard to trust the company, even as it continues to say it doesn’t factor AMP into results.
The true irony of AMP is that even as publishers are jumping off the platform, many also acknowledge that, actually, AMP is pretty good now. It supports comments and more interactive elements; it’s still fast and simple. Now that it’s run by the OpenJS Foundation and separated from the search results incentive, it appears to be on track to become a genuinely useful project. It’s not likely to replace HTML anytime soon, but it could help usher in the idea of portable and embeddable content that Jarvis and Gingras imagined all those years ago. Developers can even use AMP to make web-based projects that feel like Instagram Stories or the TikTok feed. “AMP potentially could have been — in some ways, I still think possibly could be — a really interesting way of syndicating content that takes that middle person out of the mix,” Pilhofer says.
Everyone I spoke to also thinks Core Web Vitals is a good and valuable idea, too. Speed matters more than ever; how you hit the mark doesn’t matter as much.
One source I spoke to wondered aloud if the internet might be a different place if the first versions of AMP had actually been good. Would publishers have thrown even more resources into supporting the format, giving Google even more control over how the web works — and, as the antitrust lawsuits allege, how it makes money? It certainly seems possible.
But one thing proved undeniable: for Google, there was simply no coming back from the first days of AMP, when publishers felt like the company was making grand pronouncements about saving the web while also force-feeding them bad products that served Google’s ends and no one else’s. Even Facebook Instant Articles and Apple News, constrained and problematic as they were, felt optional. AMP didn’t.
“It maybe had good intentions about making the mobile web better,” Adams says, “but went about it in probably one of the worst ways you could have imagined. It was a PR nightmare.”
One of the smartest and most profitable things Google ever did was align itself with the growth of the web. It offered useful free services, used projects like Fiber and Android to help get more people online, and made the sprawling internet a little easier for people to navigate. As the web grew, so did Google, both to great heights. But when the web was threatened by the rise of closed platforms, Google mortgaged many of its ideas about openness in order to make sure the profits kept coming. “And as a long-term effect, it probably woke a lot of news publishers up to the fact that Google is maybe not a benign entity,” Adams says. “And we need to take their dominance a bit more seriously as a news story in its own right.”
In response to this story, Google spokesperson Meghann Farnsworth said the company “will continue to collaborate with the industry to build technology that provides helpful experiences for users, delivers value to publishers and creators and helps contribute to a healthy ecosystem and the open web.”
Google is still the web’s biggest and most influential company. But across the publishing industry, it’s no longer seen as a partner. AMP ultimately neither saved nor killed the open web. But it did kill Google’s good name — one not-that-fast webpage at a time.
Casey Newton and Nilay Patel contributed reporting.
